AI SuperHero Intelligence Division
Lvl 1 · Rookie

AI SuperHero

No drawing skills required.

Visual Conjuring.

Turn words into pictures. Create images, logos and thumbnails without drawing a single line.

BriefingPower 02 of 06

Describe it. See it.

An image generator turns a written description into a picture in a few seconds. If you can describe a photo to a friend over the phone, you can do this.

You don't need special software to begin. ChatGPT, Gemini and Copilot can all create images inside the chat window. Type what you want, look at the result, then ask for changes in plain words.

Good first uses

  • A thumbnail for a video or post5 mins
  • An illustration for a slide or flyer5 mins
  • Rough logo ideas to brief a designer10 mins
  • A birthday card for someone hard to buy for5 mins

The skill is all in the describing. Most poor images come from prompts that name a subject and nothing else.

Field techniqueLearn this once

The image recipe.

Four ingredients. Use all four and you'll get close on the first try.

1. SubjectA ginger cat wearing a tiny red cape 2. Styleflat cartoon illustration, bold outlines 3. Settingon a rooftop at sunset, warm light 4. Shapewide 16:9, space on the left for a title

One prompt, four parts. Change any one part and you get a different picture.

You wantStyle words to tryGood for
A photo lookphotograph, natural light, shallow depth of fieldBlog posts, adverts
A drawn lookflat illustration, line art, watercolour, comic styleSlides, thumbnails, cards
A logo ideasimple vector logo, two colours, plain white backgroundBriefing a designer
A 3D look3D render, soft studio lighting, clay styleProduct mock-ups, icons

Edit by talking

Don't start again when it's nearly right. Say what to change: "same image, but make it night-time", "remove the text", "make the cat bigger".

Training missionsThree missions, about 10 minutes each

Your training missions.

Use the image feature in your chat assistant. Replace the parts in [brackets].

  1. Mission 1: your first image. Use all four ingredients.
    Mission 1
    Create an image of [subject, with one or two details].
    Style: [e.g. flat cartoon illustration with bold outlines].
    Setting: [place, time of day, lighting].
    Shape: [square / wide 16:9 / tall 9:16].
  2. Mission 2: a thumbnail with room for a title. Leave the words off and add them yourself afterwards. They'll be sharper.
    Mission 2
    Create a wide 16:9 thumbnail image about [topic]. Bright colours, one clear subject on the right-hand side, plain uncluttered space on the left where I can add a title later. No text in the image.
  3. Mission 3: three logo directions. Treat these as ideas to react to, not a finished logo.
    Mission 3
    Show me three different simple logo concepts for [business name], a [type of business]. Two colours at most, plain white background, no small details. Make each concept clearly different in style.
Threat assessmentWatch-outs

Villains to watch.

Wonky words and hands. Text inside images is often misspelt, and fingers still go wrong. Zoom in before you use anything.

Other people's property. Don't ask for real people, famous characters or another company's logo. Describe your own idea instead.

The small print. If the image is for something you'll sell or advertise, check the tool's terms on commercial use first. They differ between tools and plans.

Rule of thumb: AI images are a fast first draft. For a logo you'll put on a van, use them to brief a designer.

DebriefPower 02 complete when

You've got this power when...

  • You include subject, style, setting and shape
  • You edit by conversation instead of starting again
  • You add titles yourself instead of asking the AI to spell them
  • You check the terms before using an image commercially

Next: Word Forging